CN107707497B - Communication signal identification method based on subtraction clustering and fuzzy clustering algorithm - Google Patents

Communication signal identification method based on subtraction clustering and fuzzy clustering algorithm Download PDF

Info

Publication number
CN107707497B
CN107707497B CN201710319971.6A CN201710319971A CN107707497B CN 107707497 B CN107707497 B CN 107707497B CN 201710319971 A CN201710319971 A CN 201710319971A CN 107707497 B CN107707497 B CN 107707497B
Authority
CN
China
Prior art keywords
clustering
fuzzy
fuzzy clustering
constellation
value
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201710319971.6A
Other languages
Chinese (zh)
Other versions
CN107707497A (en
Inventor
邵怀宗
肖恒
王文钦
潘晔
陈慧
胡全
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
University of Electronic Science and Technology of China
Original Assignee
University of Electronic Science and Technology of China
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by University of Electronic Science and Technology of China filed Critical University of Electronic Science and Technology of China
Priority to CN201710319971.6A priority Critical patent/CN107707497B/en
Publication of CN107707497A publication Critical patent/CN107707497A/en
Application granted granted Critical
Publication of CN107707497B publication Critical patent/CN107707497B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L27/00—Modulated-carrier systems
    • H04L27/0012—Modulated-carrier systems arrangements for identifying the type of modulation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques

Abstract

The invention discloses a communication signal identification method based on subtraction clustering and fuzzy clustering algorithms, which comprises initialization parameters; setting different initial field radius values aiming at different subcarriers, and clustering constellation points of the received communication signals by adopting a subtractive clustering algorithm; when the number of the subtractive clustering centers is smaller than a first preset threshold value, reducing the radius of the neighborhood, and continuing the subtractive clustering; clustering constellation points of the communication signals again by adopting a fuzzy clustering algorithm by taking the subtraction clustering centers with higher density of a first preset threshold value as initial centers of the fuzzy clustering algorithm; designating the initial clustering number of fuzzy clustering, and evaluating the rationality of clustering by combining the Xie-Beni index and the relative radius of a clustered constellation map, wherein if the initial clustering number is unreasonable, iteration is needed; comparing the relative radius with the radius of the standard constellation diagram, the modulation mode of the signal can be obtained, and the type of the standard modulation signal is the type of the communication signal.

Description

Communication signal identification method based on subtraction clustering and fuzzy clustering algorithm
Technical Field
The invention relates to the technical field of communication, in particular to a communication signal identification method based on subtraction clustering and fuzzy clustering algorithms.
Background
Modulated signal identification plays a key role in a variety of civilian and military fields, such as cognitive radio, spectrum monitoring, and the like. In practical wireless communications, multipath channels can cause distortion of the signal and signal identification becomes more challenging. At present, in non-cooperative communication, when identifying OFDM signals of different standard protocols, an important step is to identify effective subcarriers of OFDM.
In the modulation identification research based on constellation clustering and neural network, subtraction clustering and fuzzy clustering are used for identifying single carrier signals, the signals can be effectively identified when SNR is 15dB, but the field radius value in the subtraction clustering is a fixed value, and when the types of the signal sets to be identified are more, the identification rate is reduced.
When the subcarrier signals are identified by using subtractive clustering in the research of a high-order QAM signal modulation identification algorithm under a multipath fading environment, the signal-to-noise ratio is introduced into the field radius to be suitable for signal constellation points with different densities, but in non-cooperative communication, the signal-to-noise ratio is actually unknown and needs to be estimated, so that the calculated amount of the whole algorithm is increased, errors also exist after estimation, the errors are accumulated, the identification rate is finally influenced, a correlation function is constructed by using noise power, the final clustering number is controlled by using the function, and the complexity of the whole algorithm is increased.
Disclosure of Invention
Aiming at the defects in the prior art, the communication signal identification method based on the subtractive clustering and the fuzzy clustering algorithm solves the problems of low communication signal identification rate and high complexity of the conventional clustering algorithm.
In order to achieve the purpose of the invention, the invention adopts the technical scheme that:
a communication signal identification method based on subtraction clustering and fuzzy clustering algorithm is provided, which comprises the following steps:
initializing a domain radius, an effectiveness function variable, a convergence threshold value of a fuzzy clustering function and the maximum iteration times of the fuzzy clustering function;
clustering constellation points of the received communication signals by adopting a subtractive clustering algorithm, and outputting a plurality of subtractive clustering centers;
when the number of the subtraction clustering centers is smaller than a first preset threshold value, the radius of the field is reduced according to a second preset threshold value, and the subtraction clustering algorithm is adopted again to cluster the constellation points of the communication signals until the number of the subtraction clustering centers is larger than or equal to the first preset threshold value;
using a first preset threshold value subtraction clustering centers with higher density in the subtraction clustering centers as initial centers of a fuzzy clustering algorithm, clustering constellation points of communication signals by adopting the fuzzy clustering algorithm, and outputting a plurality of obtained fuzzy clustering centers;
calculating the distance of each fuzzy clustering center relative to the origin of the constellation diagram, arranging all the distances in a descending order, and calculating the relative radius of the communication signal constellation diagram by adopting the distance of the first half part and the distance of the second half part;
and searching a standard radius value corresponding to the standard modulation signal constellation diagram with the same clustering center number through the fuzzy clustering center number corresponding to the relative radius, wherein when the difference between the relative radius and the standard radius value is smaller than a third preset threshold value, the category of the standard modulation signal is the category of the communication signal.
The invention has the beneficial effects that: the initial field radius of different communication signals during subtraction clustering can be dynamically determined, and when the number of subtraction clustering centers is smaller than a first preset threshold value, the field radius is reduced according to a second preset threshold value, so that the field radius values corresponding to the different modulation signals can be obtained, and the obtained clustering number and the subtraction clustering centers are more accurate.
In the process of identifying the communication signals, a subtraction clustering center obtained by a subtraction clustering algorithm is used as an initial center of fuzzy clustering, so that the fuzzy clustering can be used in blind signal identification, the specified process of updating the clustering number is combined with the process of updating the field radius, accurate identification of most communication signals can be realized, and the accuracy and stability of the obtained clustering number and the clustering center are far higher than those of a single clustering algorithm through a simulation test.
The subtraction clustering center obtained after subtraction clustering is used for replacing the initial clustering center of the fuzzy clustering, so that the iteration times of the fuzzy clustering can be reduced, and the center obtained by subtraction clustering is corrected again by using the fuzzy clustering, so that the obtained clustering center is more reliable.
Drawings
Fig. 1 is a flow chart of an embodiment of a communication signal identification method based on subtractive clustering and fuzzy clustering algorithms.
Fig. 2a is a simulation diagram of obtaining a subcarrier BPSK signal clustering center by using subtractive clustering, fuzzy clustering and the method of the present embodiment.
Fig. 2b is a simulation diagram of obtaining a subcarrier QPSK signal clustering center by using subtractive clustering, fuzzy clustering and the method of the present embodiment.
Fig. 2c is a simulation diagram of the method for obtaining the subcarrier 16QAM signal clustering center by adopting subtraction clustering, fuzzy clustering and the present solution.
Fig. 2d is a simulation diagram of the method for obtaining the subcarrier 64QAM signal clustering center by adopting subtraction clustering, fuzzy clustering and the present solution.
Detailed Description
The following description of the embodiments of the present invention is provided to facilitate the understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and it will be apparent to those skilled in the art that various changes may be made without departing from the spirit and scope of the invention as defined and defined in the appended claims, and all matters produced by the invention using the inventive concept are protected.
Referring to fig. 1, fig. 1 shows a flow chart of an embodiment of a communication signal identification method based on subtractive clustering and fuzzy clustering algorithms, as shown in fig. 1, the method 100 comprises steps 101 to 106.
In step 101, a domain radius, an effectiveness function variable, a convergence threshold of the fuzzy clustering function, and a maximum iteration number of the fuzzy clustering function are initialized. In specific implementation, the variable OldF of the validity function is 10e10, the convergence threshold epsilon of the fuzzy clustering function is 1e-5, and the maximum iteration number max _ iter of the fuzzy clustering function is 100.
In step 102, a subtractive clustering algorithm is used to cluster the constellation points of the received communication signal, and a plurality of subtractive clustering centers are output.
In an embodiment of the present invention, the clustering the constellation points of the received communication signal by using a subtractive clustering algorithm further includes:
calculating the density value of each constellation point in the communication signal:
Figure BDA0001289515100000041
wherein D isiIs a density value; gamma rayaIs the radius of the field; x is the number ofjAnd xiThe constellation points are jth and ith constellation points in the communication signal; n is the total number of constellation points in the communication signal; exp is the index of e; | xi-xjThe | | is the distance between the ith constellation point and the jth constellation point;
and taking the maximum density value as a subtraction clustering center, and calculating the density values of the remaining constellation points in the communication signal:
Figure BDA0001289515100000042
wherein the content of the first and second substances,
Figure BDA0001289515100000043
is the maximum value in the density values of the last constellation points,
Figure BDA0001289515100000044
the constellation point corresponding to the maximum value in the density value of the last constellation point;
Figure BDA0001289515100000045
is xiAnd
Figure BDA0001289515100000046
the distance between them; gamma rayb＝1.25γa；
And when the density values of the obtained subtraction clustering centers which can cover all the constellation points or the rest constellation points are smaller than a preset value, outputting all the obtained subtraction clustering centers.
Multiple simulation experiments show that in a Gaussian channel, when the SNR is 13dB and the subcarrier is a BPSK signal, when gamma is the BPSK signalaWhen the signal is approximately equal to 0.35, the clustering effect is best (the subtraction clustering center is closer to the standard center of the original signal, and the clustering number is closer to the optimal clustering number); similarly, when γ isaWhen the signal is approximately equal to 0.3, the clustering effect of the QPSK signal is the best; when gamma isaWhen the signal is approximately equal to 0.2, the clustering effect of the 16QAM signal is the best; when gamma isaWhen the amplitude is approximately equal to 0.1, the clustering effect of the 64QAM signal is the best.
In step 103, when the number of the subtractive clustering centers is smaller than the first preset threshold, the radius of the domain is decreased according to the second preset threshold, and the subtractive clustering algorithm is used to cluster the constellation points of the communication signals again until the number of the subtractive clustering centers is greater than or equal to the first preset threshold.
In practice, the second preset threshold value of the field radius reduction according to the second preset threshold value is preferably 0.01, and the updated field radius is new gammaa＝γa-0.01. The first predetermined threshold is preferably 2, which is actually the initial cluster center number.
In step 104, the subtraction clustering centers with the first preset threshold value with higher density in the subtraction clustering centers are used as initial centers of the fuzzy clustering algorithm, the constellation points of the communication signals are clustered by the fuzzy clustering algorithm, and the obtained plurality of fuzzy clustering centers are output.
In step 103, assuming that the first preset threshold is 4, and the number of the clustering centers after the initial clustering is less than 4, the neighborhood radius is reduced by 0.01, and the subtraction clustering algorithm is adopted to cluster the constellation points of the communication signals again until the number of the subtraction clustering centers is greater than or equal to 4, because the subtraction clustering is performed automatically, 1, 2, 3 or more subtraction clustering centers may be added once per clustering, that is, the number added each time is indefinite, and assuming that the number of the subtraction clustering centers after the clustering is completed is 6, in step 104, when the fuzzy clustering algorithm is adopted, the first 4 subtraction clustering centers with higher density in the subtraction clustering centers are used as initial centers of the fuzzy clustering algorithm.
In an embodiment of the present invention, the clustering the constellation points of the communication signal by using the fuzzy clustering algorithm further includes:
calculating a fuzzy clustering center of a fuzzy clustering algorithm according to the initial centers of the first preset threshold value:
Figure BDA0001289515100000061
wherein v isiIs a fuzzy clustering center; n is the total number of constellation points in the communication signal; mu.sijIs a constellation point xjIs classified into a fuzzy clustering center viDegree of membership of 0. ltoreq. mu.ijLess than or equal to 1; m is a fuzzy factor of the fuzzy clustering algorithm, and m belongs to [1, ∞);
calculating a cost function value of a fuzzy clustering algorithm:
Figure 1
wherein d isij＝||vi-xj| |, which is the fuzzy clustering center viAnd constellation point xjThe distance between them; m is a fuzzy factor, and m belongs to [1, ∞); lambda [ alpha ]j J 1.. and n is lagrange factor;
when the cost function value is smaller than the convergence threshold value of the fuzzy clustering function, or the iteration times of calculating the fuzzy clustering center is smaller than the maximum iteration times of the fuzzy clustering function, updating the membership degree of the constellation point classified to each fuzzy clustering center:
Figure BDA0001289515100000063
wherein d iskj＝||vk-xj| |, which is the fuzzy clustering center vkAnd constellation point xjThe distance between them;
calculating the fuzzy clustering center of the fuzzy clustering algorithm by adopting the membership degree of each fuzzy clustering center to which the updated constellation point is classified;
and when the calculated cost function value is larger than or equal to the convergence threshold value of the fuzzy clustering function, or the iteration number of the calculated fuzzy clustering center is equal to the maximum iteration number of the fuzzy clustering function, outputting the membership degree of the fuzzy clustering center and each constellation point classified to each fuzzy clustering center.
Multiple times of simulation shows that under a Gaussian channel, when the SNR is 13dB, in the process of fuzzy clustering of BPSK signals, QPSK signals, 16QAM signals and 64QAM signals, when m is approximately equal to 2, the obtained clustering center is closer to the standard center of the original signal relative to other fuzzy factor m values, the obtained clustering number is accurate, and the four subcarrier signals have good clustering effect.
In step 105, calculating the distance of each fuzzy cluster center relative to the origin of the constellation diagram, arranging all the distances in a descending order, and calculating the relative radius of the communication signal constellation diagram by adopting the first half distance and the second half distance;
in one embodiment of the present invention, calculating the relative radius of the communication signal constellation using the first-half distance and the second-half distance further comprises:
when the number of the fuzzy clustering centers is more than or equal to a preset number (when the implementation is carried out, the preset number is preferably 4), the ratio of the average value of the distance of the front preset number in the distance of the front half part to the average value of the distance of the front preset number in the distance of the rear half part is used as the relative radius of the communication signal constellation diagram;
and when the number of the fuzzy clustering centers is less than the preset number, taking the ratio of the average value of the distance of the first half part to the average value of the distance of the second half part as the relative radius of the communication signal constellation.
The relative radius acquired by adopting the mode can further ensure the accuracy of the relative radius, and further ensure the accuracy rate in the process of identifying the communication signal.
In step 106, a standard radius value corresponding to the standard modulation signal constellation diagram with the same number of cluster centers is searched for by the fuzzy cluster center number corresponding to the relative radius, and when the difference between the relative radius and the standard radius value is smaller than a third preset threshold, the category of the standard modulation signal is the category of the communication signal.
In an embodiment of the present invention, the outputting the obtained plurality of fuzzy cluster centers and the calculating the distance from each fuzzy cluster center to the origin of the constellation further includes:
and classifying the constellation points output by the fuzzy clustering algorithm to the membership degree of each fuzzy clustering center, and calculating a clustering effectiveness function value.
In practice, the following example may be used to calculate the cluster validity function value:
Figure BDA0001289515100000081
wherein x isjIs the jth constellation point in the communication signal; c is the number of fuzzy clustering centers; v. ofiIs the ith fuzzy clustering center; n is the total number of constellation points in the communication signal; mu.sijIs a constellation point xjIs classified into a fuzzy clustering center viDegree of membership of; when u isij＞ukjWhen is delta ij1, otherwise, δij＝0；μkjIs a constellation point xjIs classified into a fuzzy cluster center xkThe membership degree of (k) is not equal to i; i.e. | is distance solving operation; | xj-vi| | is the jth fuzzy clustering center vjAnd the ith constellation point xiThe distance between them; | v | (V)i-vj| | is the ith fuzzy clustering center viWith the jth fuzzy clustering center vjThe distance between them.
Increasing the first preset threshold value according to a set multiple, then updating the first preset threshold value, decreasing the radius of the field according to a second preset threshold value, then updating the radius of the field, and updating the effectiveness function variable by adopting the clustering effectiveness function value; in practice, the setting multiple is preferably 2, and the second predetermined threshold value and the aforementioned second predetermined threshold value are the same parameter, which is preferably 0.01.
And clustering the constellation points of the communication signals by adopting a subtractive clustering algorithm according to the updated first preset threshold, the updated field radius and the updated effectiveness function variable until the effectiveness function value is greater than the effectiveness function variable.
When evaluating the effectiveness of fuzzy clustering, in general, the smallest cluster effectiveness function value F corresponds to the best number of clusters c. However, in signal recognition, it is sometimes the case that, although the F value obtained when c is 4 is smaller than c 8, the smallest F value may occur when c is 7 or c is 3 because M of the signal is an even number.
The invention can greatly improve the optimal number of fuzzy clustering centers through the clustering effectiveness function value F and the relative radius after the constellation diagram clustering, thereby ensuring the accuracy of judging the adjustment type of the communication signal.
In implementation, in this scheme, preferably, when the updated first preset threshold is greater than or equal to the fourth preset valve (preferably 8 here), the method further includes calculating a relative radius of the communication signal constellation when the fuzzy clustering center is one-fourth of the first preset threshold and one-half of the first preset threshold;
and when the updated first preset threshold value is more than or equal to one half of the fourth preset valve and less than the fourth preset valve, calculating the relative radius of the communication signal constellation when the fuzzy clustering center is one half of the first preset threshold value.
Because the first preset threshold value is updated according to the set multiple, the subtraction cluster centers and the fuzzy cluster centers with smaller number are calculated before, and the relative radius can be calculated only by directly calling the corresponding number of fuzzy cluster centers acquired before.
When there are multiple relative radii, the standard radius values corresponding to the number of the multiple fuzzy clustering centers are read in step 106, and the category of the communication signal is judged according to the relative radius that can satisfy the condition that the difference is smaller than the third preset threshold. If the plurality of relative radii all meet the condition that the difference is smaller than the third preset threshold value, the relative radius corresponding to the smallest difference can be selected through the minimum distance principle to judge the category of the communication signal.
According to the scheme, the communication signals can select OFDM effective subcarrier signals, the OFDM effective subcarrier signals comprise MPSK signals and MQAM signals, and the initial field radiuses of the MPSK signals and the MQAM signals are different.
The following describes the effect of the communication signal identification method of the present embodiment with reference to simulation:
when simulation is carried out, the OFDM signal selects a DVB-T standard, a mode selects 2K-point FFT, the number of subcarriers is 2048, the duration of a symbol is 280us, a guard interval is 56us, a sampling frequency is 10MHz, a carrier frequency is 2MHz, the modulation modes of the subcarriers are S ═ BPSK, QPSK,16QAM and 64QAM, the phase of the QPSK is integral multiple of pi/2, the number of OFDM symbols when the subcarriers are 64QAM is 5000, the number of symbols of other three subcarriers is 2000, and the signal-to-noise ratio is 13 dB. After the received signal is de-CP and FFT transformed, one of the effective sub-carriers is randomly extracted for identification, wherein the sub-carrier with the sequence number of 300 is extracted.
And under a Gaussian channel, comparing the performances of subtraction clustering (algorithm one), fuzzy clustering (algorithm two) and the method (algorithm three) for identifying the signal types aiming at the subcarrier 16QAM signals.
When the relative radius R value is calculated, the mean value of the R obtained by the third algorithm is closest to the standard radius value, the variance of the mean value is also minimum, and the second algorithm is next. It can be shown that algorithm three is the best in terms of accuracy of estimating the relative radius value or stability, i.e. algorithm three indeed improves the clustering performance of algorithm one. When both algorithm two and algorithm three are in the optimal cluster number, the iteration number n of algorithm three is smaller than that of algorithm two in terms of mean and variance. It can be shown that algorithm three actually shortens the number of iterations of algorithm two.
Table 1 table for comparing clustering performance of subcarrier 16QAM by three algorithms under gaussian channel
Figure BDA0001289515100000101
As shown in FIGS. 2a and 2b, for Gaussian channel, when the sub-carriers are BPSK and QPSK signals, the c (cluster center, all c mentioned in this application represent cluster center) after clustering by three algorithms is the same, and as can be seen from FIG. 2a, algorithm one is on the same phase axis [ -1, -0.995]Orthogonal axis [ -5,5 [ ]]*10-3The distance between the cluster center obtained in the region and the standard center is farther.
As can be seen from FIG. 2c, for a 16QAM signal, algorithm two is on the in-phase axis [0.3,0.4 ]]Orthogonal axis [ -0.9, -1.05 [)]Two cluster centers are obtained in the region, because the initial membership matrix is randomly given, and the third algorithm avoids the situation because the radius gamma of the field of the subtractive clustering isaAnd gammabThe arrangement is reasonable, so that the distances between the obtained clustering centers are not very close to each other.
As can be seen from fig. 2d, when the subcarrier is 64QAM, the number of clusters finally clustered by the algorithm one because the radius of the domain is not suitable at this moment, and the reason that the algorithm three can avoid the radius of the domain is a fine-tuning and iterative process, new γ, is 27a＝γa0.01, and a joint evaluation of the validity function and relative radius on the validity of the clustering.
As can be seen from the simulation of the signal in fig. 4, the signal modulation order and the cluster center obtained by the third algorithm are the best compared with those of the other two algorithms.
Table 2 shows the modulation order M and the relative radius R obtained by clustering different subcarrier signals by the three algorithms. It can be seen that in the gaussian channel, when the SNR is 13dB, the subcarriers are BPSK and QPSK signals, and the identification accuracy of the three algorithms is equivalent, but the algorithm is poor in estimating the accuracy on the relative radius. The subcarrier is a 16QAM signal, the relative radius after the algorithm is clustered for three times is closer to the standard center, and the algorithm is performed for one time, which shows that the radius value of the current field of the algorithm for one time is set more reasonably. The subcarrier is a 64QAM signal, and the calculated relative radius value result is poor due to the fact that identification errors occur during simulation of the first algorithm, and the relative radius value of the third algorithm is estimated more accurately relative to the relative radius value of the second algorithm.
Table 2M, R parameter table for clustering different sub-carriers by three algorithms under gaussian channel
Figure BDA0001289515100000111
As can be seen from table 3, in the gaussian channel, when the SNR is 13dB, when the subcarrier is a BPSK signal in the clustering process of algorithm three, the F value at c-2 is smaller than that at c-4, and the R value corresponding thereto is closer to the standard radius value of the cluster number corresponding thereto; when the subcarrier is a QPSK signal, the F value at c-8 is greater than that at c-4, indicating that the range of c for the smallest F value should be between [2,8], and the R value at c-4 is closer to the standard radius value of its corresponding cluster number; similarly, when the subcarrier is a 16QAM signal, c is more than or equal to 8 and less than or equal to 32 after the range is narrowed by using the F value, and the R value at the position where c is 16 is closer to the standard radius value of the corresponding cluster number; when the subcarrier is a 64QAM signal, c is more than or equal to 64 and less than or equal to 256 after the range is narrowed down by using the F value, and the R value at the position where c is 64 is more close to the standard radius value of the corresponding cluster number.
TABLE 3C, F, R parameter table for clustering three pairs of different sub-carriers in algorithm under Gaussian channel
Figure BDA0001289515100000121

Claims (8)

1. The communication signal identification method based on the subtractive clustering and the fuzzy clustering algorithm is characterized by comprising the following steps:
initializing a domain radius, an effectiveness function variable, a convergence threshold value of a fuzzy clustering function and the maximum iteration times of the fuzzy clustering function;
clustering constellation points of the received communication signals by adopting a subtractive clustering algorithm, and outputting a plurality of subtractive clustering centers;
when the number of the subtraction clustering centers is smaller than a first preset threshold value, the radius of the field is reduced according to a second preset threshold value, and the subtraction clustering algorithm is adopted again to cluster the constellation points of the communication signals until the number of the subtraction clustering centers is larger than or equal to the first preset threshold value;
using a first preset threshold value subtraction clustering centers with higher density in the subtraction clustering centers as initial centers of a fuzzy clustering algorithm, clustering constellation points of communication signals by adopting the fuzzy clustering algorithm, and outputting a plurality of obtained fuzzy clustering centers;
calculating the distance of each fuzzy clustering center relative to the origin of the constellation diagram, arranging all the distances in a descending order, and calculating the relative radius of the communication signal constellation diagram by adopting the distance of the first half part and the distance of the second half part;
and searching a standard radius value corresponding to the standard modulation signal constellation diagram with the same clustering center number through the fuzzy clustering center number corresponding to the relative radius, wherein when the difference between the relative radius and the standard radius value is smaller than a third preset threshold value, the category of the standard modulation signal is the category of the communication signal.
2. The method of claim 1 wherein computing the relative radius of the communication signal constellation using the first-half distance and the second-half distance further comprises:
when the number of the fuzzy clustering centers is larger than or equal to the preset number, taking the ratio of the average value of the distance of the front preset number in the distance of the front half part to the average value of the distance of the front preset number in the distance of the rear half part as the relative radius of the communication signal constellation;
and when the number of the fuzzy clustering centers is less than the preset number, taking the ratio of the average value of the distance of the first half part to the average value of the distance of the second half part as the relative radius of the communication signal constellation.
3. The method for identifying a communication signal based on subtractive clustering and fuzzy clustering according to claim 1 or 2, wherein said outputting a plurality of fuzzy cluster centers and said calculating a distance between each fuzzy cluster center and a constellation origin further comprises:
classifying the constellation points output by the fuzzy clustering algorithm to the membership degree of each fuzzy clustering center, and calculating a clustering effectiveness function value;
increasing the first preset threshold value according to a set multiple, then updating the first preset threshold value, decreasing the radius of the field according to a second preset threshold value, then updating the radius of the field, and updating the effectiveness function variable by adopting the clustering effectiveness function value;
and clustering the constellation points of the communication signals by adopting a subtractive clustering algorithm according to the updated first preset threshold, the updated field radius and the updated effectiveness function variable until the effectiveness function value is greater than the effectiveness function variable.
4. The method of claim 3, wherein when the updated first predetermined threshold is greater than or equal to the fourth predetermined threshold, the method further comprises calculating the relative radius of the communication signal constellation when the fuzzy clustering center is one-fourth of the first predetermined threshold and one-half of the first predetermined threshold;
and when the updated first preset threshold value is more than or equal to one half of the fourth preset valve and less than the fourth preset valve, calculating the relative radius of the communication signal constellation when the fuzzy clustering center is one half of the first preset threshold value.
5. The method for identifying a communication signal based on subtractive clustering and fuzzy clustering according to claim 4, wherein said specific formula for calculating the cluster validity function value is:
Figure FDA0002389881200000021
wherein x isjIs the jth constellation point in the communication signal; c is the number of fuzzy clustering centers; v. ofiIs the ith fuzzy clustering center; n is the total number of constellation points in the communication signal; mu.sijIs a constellation point xjIs classified into a fuzzy clustering center viDegree of membership of; when u isij>ukjWhen is deltaij1, otherwise, δij＝0；μkjIs a constellation point xjIs classified into a fuzzy cluster center xkThe membership degree of (k) is not equal to i; i.e. | is distance solving operation; | xj-vi| | is the jth fuzzy clustering center vjAnd the ith constellation point xiThe distance between them; | v | (V)i-vj| | is the ith fuzzy clustering center viWith the jth fuzzy clustering center vjThe distance between them.
6. The method of claim 1, 2, 4 or 5, wherein the clustering constellation points of the received communication signal using the subtractive clustering algorithm further comprises:
calculating the density value of each constellation point in the communication signal:
Figure FDA0002389881200000031
wherein D isiIs a density value; gamma rayaIs the radius of the field; x is the number ofjAnd xiThe constellation points are jth and ith constellation points in the communication signal; n is the total number of constellation points in the communication signal; exp is the index of e; | xi-xjThe | | is the distance between the ith constellation point and the jth constellation point;
and taking the maximum density value as a subtraction clustering center, and calculating the density values of the remaining constellation points in the communication signal:
Figure FDA0002389881200000032
wherein the content of the first and second substances,
Figure FDA0002389881200000033
is the maximum value in the density values of the last constellation points,
Figure FDA0002389881200000034
the constellation point corresponding to the maximum value in the density value of the last constellation point;
Figure FDA0002389881200000035
is xiAnd
Figure FDA0002389881200000036
the distance between them; gamma rayb＝1.25γa；
And when the density values of the obtained subtraction clustering centers which can cover all the constellation points or the rest constellation points are smaller than a preset value, outputting all the obtained subtraction clustering centers.
7. The method of claim 6, wherein clustering constellation points of the communication signal using the fuzzy clustering algorithm further comprises:
calculating a fuzzy clustering center of a fuzzy clustering algorithm according to the initial centers of the first preset threshold value:
Figure FDA0002389881200000041
wherein v isiIs a fuzzy clustering center; n is the total number of constellation points in the communication signal; mu.sijIs a constellation point xjIs classified into a fuzzy clustering center viDegree of membership of 0. ltoreq. mu.ijLess than or equal to 1; m is a fuzzy factor of the fuzzy clustering algorithm, and m belongs to [1, ∞);
calculating a cost function value of a fuzzy clustering algorithm:
Figure FDA0002389881200000042
wherein d isij＝||vi-xj| |, which is the fuzzy clustering center viAnd constellation point xjThe distance between them; m is a fuzzy factor, and m belongs to [1, ∞); lambda [ alpha ]jJ 1.. n is lagrange's factorA seed;
when the cost function value is smaller than the convergence threshold value of the fuzzy clustering function, or the iteration times of calculating the fuzzy clustering center is smaller than the maximum iteration times of the fuzzy clustering function, updating the membership degree of the constellation point classified to each fuzzy clustering center:
Figure FDA0002389881200000043
wherein d iskj＝||vk-xj| |, which is the fuzzy clustering center vkAnd constellation point xjThe distance between them; c is the number of fuzzy clustering centers;
calculating the fuzzy clustering center of the fuzzy clustering algorithm by adopting the membership degree of each fuzzy clustering center to which the updated constellation point is classified;
and when the calculated cost function value is larger than or equal to the convergence threshold value of the fuzzy clustering function, or the iteration number of the calculated fuzzy clustering center is equal to the maximum iteration number of the fuzzy clustering function, outputting the fuzzy clustering center and the membership degree of each constellation point classified to each fuzzy clustering center.
8. The method for identifying communication signals based on subtractive clustering and fuzzy clustering according to claim 1, 2, 4, 5 or 7, wherein the communication signals are OFDM effective subcarrier signals, and the OFDM effective subcarrier signals comprise MPSK signals and MQAM signals; the MPSK signal and the MQAM signal have different initial domain radiuses.
CN201710319971.6A 2017-05-09 2017-05-09 Communication signal identification method based on subtraction clustering and fuzzy clustering algorithm Active CN107707497B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710319971.6A CN107707497B (en) 2017-05-09 2017-05-09 Communication signal identification method based on subtraction clustering and fuzzy clustering algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710319971.6A CN107707497B (en) 2017-05-09 2017-05-09 Communication signal identification method based on subtraction clustering and fuzzy clustering algorithm

Publications (2)

Publication Number Publication Date
CN107707497A CN107707497A (en) 2018-02-16
CN107707497B true CN107707497B (en) 2020-06-02

Family

ID=61169576

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710319971.6A Active CN107707497B (en) 2017-05-09 2017-05-09 Communication signal identification method based on subtraction clustering and fuzzy clustering algorithm

Country Status (1)

Country Link
CN (1) CN107707497B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109274626B (en) * 2018-11-21 2020-11-13 电子科技大学 Modulation identification method based on constellation diagram orthogonal scanning characteristics
CN109978006B (en) * 2019-02-25 2021-02-19 北京邮电大学 Face image clustering method and device
CN112134818A (en) * 2020-09-23 2020-12-25 青岛科技大学 Underwater sound signal modulation mode self-adaptive in-class identification method
CN112565126B (en) * 2020-11-26 2022-06-28 杭州电子科技大学 Modulation identification method based on receiver multi-dimensional characteristic analog-digital mixed signal

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2008022435A1 (en) * 2006-08-21 2008-02-28 Kaveh Seyed Momen Method, system and apparatus for real-time classification of muscle signals from self-selected intentional movements
CN102065056A (en) * 2011-01-10 2011-05-18 郑州大学 Method for realizing MQAM (Multiple Quadrature Amplitude Modulation) signal modulation mode identification of any constellation diagram on basis of clustering
CN103150731A (en) * 2013-03-07 2013-06-12 南京航空航天大学 Fuzzy clustering image segmenting method
CN104898562A (en) * 2015-04-27 2015-09-09 攀枝花学院 Modeling method of thermal error compensation of numerically-controlled machine tool
CN104915560A (en) * 2015-06-11 2015-09-16 万达信息股份有限公司 Method for disease diagnosis and treatment scheme based on generalized neural network clustering
CN106408569A (en) * 2016-08-29 2017-02-15 北京航空航天大学 Brain MRI (magnetic resonance image) segmentation method based on improved fuzzy C-means clustering algorithm

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2008022435A1 (en) * 2006-08-21 2008-02-28 Kaveh Seyed Momen Method, system and apparatus for real-time classification of muscle signals from self-selected intentional movements
CN102065056A (en) * 2011-01-10 2011-05-18 郑州大学 Method for realizing MQAM (Multiple Quadrature Amplitude Modulation) signal modulation mode identification of any constellation diagram on basis of clustering
CN103150731A (en) * 2013-03-07 2013-06-12 南京航空航天大学 Fuzzy clustering image segmenting method
CN104898562A (en) * 2015-04-27 2015-09-09 攀枝花学院 Modeling method of thermal error compensation of numerically-controlled machine tool
CN104915560A (en) * 2015-06-11 2015-09-16 万达信息股份有限公司 Method for disease diagnosis and treatment scheme based on generalized neural network clustering
CN106408569A (en) * 2016-08-29 2017-02-15 北京航空航天大学 Brain MRI (magnetic resonance image) segmentation method based on improved fuzzy C-means clustering algorithm

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
《Fuzzy identification based on improved clustering arithmetic and its application》;San Ye ，Ai Ling;《2011 Eighth International Conference on Fuzzy Systems and Knowledge Discovery (FSKD)》;20110915;第147-151页 *
《多径衰落环境下高阶QAM信号调制识别算法研究》;袁莹莹;《中国优秀硕士学位论文全文数据库 信息科技辑》;20150715;I136-66 *

Also Published As

Publication number Publication date
CN107707497A (en) 2018-02-16

Similar Documents

Publication Publication Date Title
CN109802905B (en) CNN convolutional neural network-based digital signal automatic modulation identification method
CN107707497B (en) Communication signal identification method based on subtraction clustering and fuzzy clustering algorithm
CN112733811B (en) Method for identifying underwater sound signal modulation modes based on improved dense neural network
CN106533590B (en) Uplink channel quality measurement method based on receiving end EVM
Zhendong et al. Modulation recognition based on constellation diagram for M-QAM signals
Zhao et al. Deep Learning-Based channel estimation with SRGAN in OFDM Systems
CN114567528B (en) Communication signal modulation mode open set recognition method and system based on deep learning
CN101036361B (en) Receiver-site restoration of clipped signal peaks
CN108696305A (en) High-precision frequency deviation measurement method suitable for LTE-A MIMO signal analysis systems
KR102370119B1 (en) Apparatus and method for detecting signals based on partial candidate in wireless communication system
CN114615122B (en) Method and device for determining frequency offset of communication signal
CN106506427B (en) A kind of STBC-OFDM Signal blind recognition method based on FOLP
CN109729035A (en) A kind of impulse noise suppression method in orthogonal frequency division multiplexi communication system
WO2011007334A2 (en) Interferer region identification using image processing
EP2140561B1 (en) A method and an apparatus for estimating a delay spread of a multipath channel
CN110166389B (en) Modulation identification method based on least square support vector machine
KR101119776B1 (en) User power offset estimation using dedicated pilot tones for ofdma
Wang et al. Low complexity Kolmogorov-Smirnov modulation classification
CN109167744B (en) Phase noise joint estimation method
KR102596571B1 (en) Deep learning-assisted index estimator for generalized LED index modulation OFDM in visible light communication
CN111935044B (en) PSK (phase Shift keying) and QAM (Quadrature amplitude modulation) signal modulation identification method based on EMD (empirical mode decomposition)
Batra et al. A Study of Automatic Modulation Classification for Rayleigh Fading Channel
Jeong et al. Channel sensing without quiet period for cognitive radio systems: A pilot cancellation approach
CN116132237B (en) OFDM signal detection and identification method based on periodic convolution and fractal
CN106817334A (en) A kind of non-linear clipping distortion compressed sensing method of estimation of multicarrier for alleviating observation interference

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant